Modeling Volatility in Biotech Stocks with a Bayesian GARCH-X Framework

Sammanfattning

This thesis examines volatility dynamics in biotechnology equities ahead of U.S Federal Drug Administration (FDA) regulatory announcements, specifically Prescription Drug User Fee Act (PDUFA) decision dates. Using a 15-year panel of daily returns from publicly traded biotech firms, we model conditional volatility with a hierarchical Bayesian GARCH-X framework incorporating market and firm-level covariates, including market capitalization, the VIX index, and a PDUFA event dummy to isolate anticipatory volatility patterns. We fit the model with Hamiltonian Monte-Carlo and then stress‑test it with posterior predictive checks and out‑of‑sample forecasts. To assess the contribution of individual volatility drivers, an Euler decomposition of Value-at-Risk (VaR) is applied to enable marginal attribution to each regressor. Further, we regress firm-specific PDUFA sensitivities on business characteristics to explore cross-sectional determinants of volatility response. Volatility rises significantly in the weeks leading up to PDUFA decisions. These findings suggest that regulatory events act as latent sources of uncertainty in biotech markets, with implications for risk management, volatility forecasting, and event-driven investment strategies in biotechnology equities.